ES2991726T3 - A replaceable bucket assembly - Google Patents

Abstract

A cuvette assembly (40) capable of being installed in an optical absorbance measurement system for measuring hemoglobin parameters in whole blood or bilirubin parameters in whole blood, the cuvette assembly (40) comprising: a cuvette substrate (41); and a cuvette module (43) fixedly connected to the cuvette substrate (41), wherein the cuvette substrate (41) is a support for securing the cuvette assembly (40) within the optical absorbance measurement system, the cuvette module (43) comprising: a sample inlet port (46); a sample outlet port (47); an electronic chip assembly (48); a sample receiving chamber (54) fluidly communicating with the sample inlet port (46) and the sample outlet port (47); a first cuvette window (49); and a second cuvette window (52) forming a portion of the sample receiving chamber (54), wherein the first cuvette window (49) and the second cuvette window (52) are aligned with each other, thereby defining a cuvette optical path length between the first cuvette window (49) and the second cuvette window (52), wherein the cuvette module (43) includes a first cuvette portion (44) and a second cuvette portion (50) joined together and thereby forming the sample receiving chamber (54), and wherein the first cuvette window (49) and the second cuvette window (52) are arranged within an optical path of the optical absorbance measuring system. (Automatic translation with Google Translate, without legal value)

Description

DESCRIPTION

A replaceable bucket assembly

Background of the invention

1. Field of the invention

The present invention relates, in general, to spectroscopic systems and methods for the identification and characterization of hemoglobin parameters in whole blood and, in particular, to a replaceable cuvette assembly capable of being installed in an optical absorbance measurement system for measuring hemoglobin parameters in whole blood or bilirubin parameters in whole blood.

2. Description of the prior art

An ultraviolet-visible spectroscopic system involves absorption spectroscopy or reflectance spectroscopy. As the name implies, these systems use light in the visible and near-ultraviolet ranges to analyze a sample. The wavelength range is typically from about 400 nm to about 700 nm. The absorption or reflectance of visible light directly affects the perceived color of the chemicals involved. UV/Vis spectroscopy is commonly used in analytical chemistry for the quantitative determination of different analytes, such as transition metal ions, highly conjugated organic compounds, and biological macromolecules. Spectroscopic analysis is commonly performed on solutions, but solids and gases can also be studied.

A near-infrared spectroscopic system also involves absorption spectroscopy or reflectance spectroscopy. These systems use light in the near-infrared range to analyze a sample. The wavelength range is typically from about 700 nm to less than 2500 nm. Typical applications include pharmaceutical, medical diagnostics (including blood sugar and pulse oximetry), food and agrochemical quality control, and combustion research, as well as research in functional neurodiagnostic imaging, sports medicine and science, elite sports training, ergonomics, rehabilitation, neonatal research, brain-computer interface, urology (bladder contraction), and neurology (neurovascular coupling).

Instrumentation for near-infrared (NIR) spectroscopy is similar to instruments for the UV-visible and mid-IR ranges. The basic parts of a spectrophotometer are a light source, a sample holder, a diffraction grating in a monochromator or prism to separate the different wavelengths of light, and a detector. The radiation source is usually a tungsten filament (300-2500 nm), a deuterium arc lamp, which is continuous over the ultraviolet region (190-400 nm), a xenon arc lamp, which is continuous from 160-2000 nm, or, more recently, light-emitting diodes (LEDs) for the visible wavelengths. The detector is usually a photomultiplier tube, a photodiode, a photodiode array, or a charge-coupled device (CCD). Single photodiode detectors and photomultiplier tubes are used with scanning monochromators, which filter the light so that only light of a single wavelength reaches the detector at a time. The scanning monochromator moves the diffraction grating to "step through" each wavelength so that its intensity can be measured as a function of wavelength. Fixed monochromators are used with CCDs and photodiode arrays. Because both devices consist of many detectors grouped into one- or two-dimensional arrays, they are capable of collecting light of different wavelengths at different pixels or groups of pixels simultaneously. Common incandescent light bulbs or quartz halogen bulbs are most often used as broadband sources of near-infrared radiation for analytical applications. Light-emitting diodes (LEDs) are also used. The type of detector used depends primarily on the wavelength range to be measured.

The main application of NIR spectroscopy in the human body utilizes the fact that the transmission and absorption of NIR light in human body tissues contains information about changes in hemoglobin concentration. By employing various wavelengths and temporally resolved (frequency or time domain) and/or spatially resolved methods, blood flow, volume, and absolute tissue saturation (StO2 or tissue saturation index (TSI)) can be quantified. Applications of NIRS oximetry include neuroscience, ergonomics, rehabilitation, brain-computer interface, urology, detection of diseases affecting blood circulation (e.g. peripheral vascular disease), detection and assessment of breast tumors, and training optimization in sports medicine.

With respect to absorption spectroscopy, the Beer-Lambert law states that the absorbance of a solution is directly proportional to the concentration of the absorbing species in the solution and the path length. Therefore, for a fixed path length, UV/Vis and NIR spectroscopy can be used to determine the concentration of the absorbing species in a solution. The method is most often used quantitatively to determine the concentrations of an absorbing species in solution, using the Beer-Lambert law:

where A is the measured absorbance, in absorbance units (AU),

It is the intensity of incident light at a given wavelength,

I is the transmitted intensity,

L the path length through the sample, and

c the concentration of the absorbing species.

For each species and wavelength, £ is a constant known as the molar absorptivity or extinction coefficient. This constant is a fundamental molecular property in a given solvent, at a given temperature and pressure, and has units of 1/M*cm or often AU/M*cm. The absorbance and extinction £ are sometimes defined in terms of the natural logarithm rather than the base 10 logarithm.

The Beer-Lambert law is useful for characterizing many compounds, but it does not hold as a universal relationship for the concentration and absorption of all substances.

Experts in the field recognize that a number of factors affect these spectroscopic systems. These factors include spectral bandwidth, wavelength error, stray light, deviations from the Beer-Lambert law, and sources of measurement uncertainty.

Stray light is an important factor affecting spectroscopic systems. Stray light causes an instrument to indicate an incorrectly low absorbance.

Deviations from the Beer-Lambert law arise as a function of concentrations. At sufficiently high concentrations, the absorption bands will saturate and show absorption flattening. The absorption peak appears to flatten because close to 100% of the light is already being absorbed. The concentration at which this occurs depends on the particular compound being measured.

Measurement uncertainty arises in quantitative chemical analysis, where results are further affected by sources of uncertainty arising from the nature of the compounds and/or solutions being measured. These include spectral interferences caused by overlapping absorption bands, colour fading of absorbing species (caused by decomposition or reaction), and possible compositional mismatch between the sample and the calibration solution.

U.S. Patent Publication No. 2003/0086074 discloses a reagentless whole blood detection system comprising a sample element in an optical path of a light source beam. In one embodiment, the sample element has a cuvette retained within a pair of channels of a housing. The cuvette comprises first and second plates having respective first and second windows that are transmissive to the light source beam and a pair of spacers that form a sample delivery passageway between the first and second plates.

Summary of the invention

Human hemoglobin (HGB) is known to be an oxygen-carrying protein in erythrocytes. Determination of its concentration in whole blood is a useful and important diagnostic tool in clinical biochemistry. COOx analyzers are used to measure blood hemoglobin parameters such as total hemoglobin (tHb), carboxyhemoglobin (COHb), deoxyhemoglobin (HHb), oxyhemoglobin (O2Hb), methemoglobin (MetHb) and fetal hemoglobin (FHb) as well as total bilirubin (tBil) using optical absorbance measurements. In practice, common COOx analyzers use lysed blood instead of whole blood due to the problems posed by spectrometric analysis of whole blood. Measurement of lysed blood is relatively simple, as the lysing process dissolves red blood cells and turns blood into a nearly non-diffusing medium. Absorbance is measured with a single collimated beam through the cuvette with little light loss due to scattering. Because of the low light loss due to scattering, a simple linear analysis can be used to find the parameters for hemoglobin and total bilirubin.

Measurement of hemoglobin and total bilirubin parameters using a whole blood sample is very difficult due to the strong optical scattering of whole blood. These problems are mainly related to handling the higher level of light scattering of whole blood compared to lysed blood. This introduces light losses and non-linear absorbance into the measurement.

The components of a prism-based spectrometer naturally have a low stray light profile. The main factor contributing to stray light performance is related to the way the components are used.

While the issues are primarily related to handling the increased level of light scattering from whole blood, there is no single factor that, if solved, is capable of solving these difficult problems. The inventors have identified several factors that need to be addressed in order to measure hemoglobin parameters in whole blood. Because whole blood is a highly diffuse medium, it is necessary to collect as much light as possible to reduce the need for an upper absorbance measurement range. It is also necessary to extend the upper limit of the measured absorbance due to the lower linearity correction range of the detector. Blood sedimentation effects are another problem that leads to poor correlation of the absorbance of whole blood scans with the absorbance of lysed blood scans. Basically, blood cells form clumps or rouleaux. The brightness of the LED white light source must also be increased. Finally, new algorithms other than those based on linearity are needed to overcome the light scattering effects of whole blood.

Typical collection optics for systems using lysed blood are designed to collect light from the cuvette in a cone approximately /-0.7 degrees wide and have an upper measurement absorbance limit of 1.5 AU (absorbance units). The inventors found that for whole blood, the system needs to collect light from the cuvette in a cone of approximately /-12 degrees and that the upper absorbance limit needed to increase to about 3.5 AU. With regard to sedimentation effects of blood, the typical time taken to measure the absorbance spectrum (approximately 1 minute), whole blood in the cuvette sediments and the blood cells form clumps or rouleaux. Consequently, scattering effects and absorbance change over time. The inventors found that changing the spectrometer control to collect multiple scans frequently rather than a few scans averaged over a longer period avoided step functions in the composite absorbance scan, which is stitched together from scans of several integration times. Unfortunately, adding more scans to extend the upper limit of absorbance increases the data collection time. To solve this dilemma, the integration time was reduced from 5 ms to 1.2 ms to reduce the data collection time. It was found, however, that this only works if the light level is increased by a corresponding factor. Therefore, the brightness of the white LED light must be increased.

Optical absorbance measurement of a diffuse sample such as whole blood presents a unique problem. The diffuse transmittance of the whole blood sample disrupts the initial spatial light distribution of the measurement system due to the typical non-uniformity of light sources. Therefore, the spatial light distribution of the "blank" scan can be very different from the whole blood sample scan. Since optical detectors have a spatially varying response, the response can vary due to changes in the spatial distribution of incident light, even though the overall intensity has not changed. An absorbance scan based on the ratio of the whole blood sample scan to the blank scan will have a significant absorbance component due to this non-uniformity of the light source, in addition to the absorbance due to the sample alone. This results in a significant measurement error of the absorbance of the whole blood sample that is intolerable for co-oximetry.

It was found that by placing the sample cuvette between diffusers, the spatial light distribution is the same for both the blank and sample scans, thus eliminating this error effect. The diffusers are specially chosen to spread an incident light beam over the entire acceptance cone of the optical system, but no more, so as to preserve the highest possible luminous flux, while dispersing the beam over the entire field.

Measurement of fetal hemoglobin parameters also presents additional challenges. These include spectral acquisition times, which must be faster. Instead of the usual 12 seconds, it must be 5 seconds or less. Spectral acquisition time includes integration time multiplied by the number of spectra combined and processing time to produce a spectrum (full light, dark, or sample) that meets all of the following requirements. Absolute wavelength accuracy must be lower; less than 0.03/-0.03 nm compared to 0.1/-0.0 nm. Wavelength calibration maintenance (less than 0.06/-0.0 nm vs. 0.1/-0.0 nm), wavelength calibration drift (less than 0.024 nm/°C vs. 0.04 nm/°C), dark current level (less than 0.06%/°C for maximum dynamic range vs. 0.1%/°C of maximum dynamic range), response nonlinearity (less than 0.06% after correction and less than 1.2% for the lowest and highest 10% of the dynamic range, compared to 0.1% after correction and 2.0% for the lowest and highest 10% of the dynamic range), straylight level (less than 0.02% of maximum dynamic range of fully illuminated detector array vs. 0.1% of maximum dynamic range of fully illuminated detector array), thermal drift of response (maximum intensity change of 10% vs. 0.1% of maximum dynamic range of fully illuminated detector array), and 6% and maximum slope of 6% over the spectral range compared to a maximum intensity change of 10% and maximum slope of 10% over the spectral range), and the allowable temperature excursion during measurement (less than 0.5°C compared to 2°C) should all be less. The present invention includes these additional features for use in measuring fetal hemoglobin parameters.

Low-cost, compact spectrometers available on the market typically use diffraction gratings (either reflective or transmissive) to disperse the input light. Diffraction gratings offer a high degree of dispersion in a small volume and produce a relatively constant bandwidth (or resolution) over the wavelength preferred by the typical user. Gratings, however, suffer from high light scattering due to multiple diffraction orders and also inherent imperfections in the lines that are etched to produce the grating surface. Therefore, mass-produced but expensive master holographic gratings are often used in applications requiring low stray light, rather than the more widely available replicated gratings.

The low stray light requirement for COOx analyzers limits the population of suitable grating manufacturers to the several around the world that produce individual precision holographic master or photo-etched gratings. This makes it difficult to obtain high performance, low cost gratings in quantity.

Prisms are also used to make spectrometers. Prisms have no problems with multiple orders of diffraction and their surfaces have orders of magnitude fewer imperfections than the surface of a grating. The components of a prism-based spectrometer naturally have a low stray light profile. Therefore, stray light in a prism spectrometer can be an order of magnitude or more lower compared to a grating spectrometer of similar design. The main factor contributing to stray light performance is the way the components are used. There are three main sources of stray light. These include (1) overfilling of the spectrometer's numerical aperture, (2) back reflection from the light array detector, and (3) focal plane imaging. Light in excess of that needed to fully illuminate the spectrometer's numerical aperture can bounce off the spectrometer and land on the detector. In the present invention, the numerical aperture of the optical fiber is 0.22 and the numerical aperture of the prism spectrometer is 0.1. A stop located over the optical fiber input restricts the light input cone of the optical fiber to prevent excess light input. The detector does not absorb all the light incident upon it, but retroreflects a portion. This retroreflection must be controlled to fall on an absorbing surface or beam trap to prevent scattering into the detector. Imparting a slight tilt to the light array detector forces the retroreflection in a harmless direction. The image of the slit in the focal plane of the detector must be as sharp as possible. Any excessive overfilling of the detector due to defocusing can be a potential source of stray light. If this light is incident on detector structures such as bond wires, metallization pads, etc., it may bounce off the sensitive surface of the detector.

Additionally, a prism spectrometer spreads the blue end of the spectrum over more pixels than a diffraction grating spectrometer and therefore the blue end of the spectrum gives a lower signal per pixel. To compensate for the lower signal per pixel, a higher-power blue LED or a cool white LED is used. The signal in the blue can be further boosted by adding an inexpensive filter glass after the LED that slightly attenuates the red end. Kopp filter glass of type 4309 glass, approximately 3 mm thick, is useful for this purpose. The main drawback of prisms is the lower dispersive power they have compared to a grating and the variation in resolution with wavelength. When a prism is used, the first disadvantage is mitigated by using a sufficiently small light array detector; the latter is mitigated because whole blood analysis does not require a uniformly small resolution throughout the waveband of interest.

Currently available spectrometers typically have a uniform resolution of 1 nm for the blood measurement spectral region of 455-660 nm. In the present disclosure, the spectral region is broadened to cover the 422-695 nm spectral region. In addition, the resolution is selectively increased in regions where low resolution is not required (such as the 600-695 nm region and the 422-455 nm region). In the present disclosure, these regions have a resolution greater than 1 nm. Typically, the resolution is about 3.0 to about 3.5 nm. These ranges are used to capture additional wavelength calibration peaks for wavelength calibration and fluid detection. The broader spectral region of the present disclosure requires consideration of the spectrum dispersed from the prism. The scattered spectrum must spread across the light array detector and cover enough pixels to sample the spectrum with fine enough resolution, but not so much as to extend outside the detector array. Because of the wider spectral range, the present disclosure incorporates a light array detector having 1024 pixels with an active area length of approximately 8.0 mm.

A minimal parts reference design for an optical dispersion spectrometer requires only two optical components: a light dispersion element (i.e. prism or grating) and a doublet (achromatic) lens. The prism/grating has a reflective coating on the base. An example of an acceptable prism is the Littrow prism. The Littrow prism has a structure such that it can be used for a compact, low cost spectrometer of the present invention. The prism material (dispersion characteristic) and the focal length of the objective are other factors to be considered. Although other prisms and achromatic lenses may be used, one embodiment of the present disclosure incorporates a Schott glass F5 prism and an 80 mm focal length lens. This particular combination provides a spectrum dispersion length of approximately 6.48 mm. This dispersion length leaves approximately 0.75 mm at each end of the light array detector available for tolerance variations and dark correction pixels.

Thermal drift of the spectral response must be taken into account. It is critical that the spectral response of the spectrometer be maintained within a certain range between full light and whole blood scans. Any change in the spectrometer response will cause absorbance errors. The primary precaution against this change is to ensure that the slit image overfills the pixels so that image drift due to temperature does not cause a reduction in light at the detector pixel. The 1:1 image of the system combined with a 200 µm diameter optical fiber fills the 125 µm high pixels. As long as the image drift is limited to less than about 30 µm of movement in any direction along the detector during a measurement interval, thermal drift is not a problem. The present disclosure also contemplates various mechanisms to minimize the effects of thermal drift on the spectral response. These mechanisms include insulating the spectrometer housing to minimize temperature changes external to the spectrometer housing, maintaining the temperature within the spectrometer housing using a temperature-controlled heat source, and/or incorporating a temperature-compensated lens mount for the achromatic lens.

Next, the process of the present disclosure that transforms the electrical signals from the spectrometer will be discussed. First, the absorbance is measured, which is minus the base ten logarithm of the ratio of the electrical signal received when the blood sample is in the cuvette to the electrical signal received when a clear fluid is in the cuvette. Second, the absorbance values at each wavelength are input into a sweep function that maps the absorbance values to analyte levels (COOx and bilirubin parameters) in the whole blood sample. The sweep function and its coefficients are established by using the measured absorbance values for whole blood samples with known analyte values, and establishing the relationship between these absorbance values and the known analyte values.

The present invention achieves these and other objectives by providing a replaceable cuvette assembly capable of being installed in a compact, low-cost COOx analyzer subsystem.

The replaceable bucket assembly comprises the functions of claim 1. Furthermore, optional functions are defined in the dependent claims.

In one embodiment of the present invention, the replaceable cuvette assembly is capable of being installed in a whole blood hemoglobin parameter measurement system including (a) a sample optical module having a light emitting module, the replaceable cuvette assembly and a calibration light module, (b) an optical fiber, (c) a spectrometer module, and (d) a processor module. The light emitting module has an LED light source capable of emitting light when directed along an optical path. The cuvette assembly is adjacent to the light emitting module, where the cuvette assembly is adapted to receive a whole blood sample and has a sample receiving chamber with a first cuvette window and a second cuvette window aligned with each other. The sample receiving chamber is disposed in the optical path to receive light from the LED light source and has an optical path length defined between the first cuvette window and the second cuvette window along with an electronic chip capable of storing a path length value of the sample receiving chamber. The calibration light module has a calibration light source with one or more known wavelengths of light where the calibration light module is capable of emitting a calibration light into the optical path. The optical fiber has a light receiving end and a light emitting end. The light receiving end is optically connected to the sample optical module, where the light receiving end receives the optical path light and conducts the light to the light emitting end. The spectrometer module receives the light from the light emitting end of the optical fiber, separates the light into a plurality of light beams where each light beam has a different wavelength and converts the plurality of light beams into an electrical signal. The processor module (1) obtains the path length value from the sample receiving chamber of the replaceable cuvette of the electronic chip and (2) receives and processes the electrical signal of the spectrometer module generated for a whole blood sample. The sample chamber path length value is used to transform the electrical signal into a usable output signal for displaying and reporting the hemoglobin and/or total bilirubin parameter values of the whole blood sample.

The light emitting module may include a plurality of optical components disposed in the optical path between the LED light source and the cuvette assembly, where the plurality of optical components includes at least one optical diffuser and one or more collimating lenses, a circular polarizer and a focusing lens.

The calibration light module may include a diffuser disposed in the optical path downstream of the cuvette assembly but upstream of a beam splitter.

In yet another aspect of the present invention, the replaceable cuvette assembly may be installed in an optical absorbance measurement system for whole blood. The system includes a sample optical module, an optical fiber, a spectrometer module, and a processor module. The sample optical module includes a light emitting module, a cuvette module, a first optical diffuser, and a second optical diffuser. The cuvette module is positioned between the first optical diffuser and the second optical diffuser. The spectrometer module receives light from the light emitting end of the optical fiber, splitting the light into a plurality of light beams and converting the plurality of light beams into an electrical signal. The processor module receives and processes the electrical signal from the spectrometer module generated for the whole blood sample and transforms the electrical signal into an output signal usable for displaying and reporting hemoglobin parameter values and/or total bilirubin parameter values for the whole blood sample.

The spectrometer module may include an entrance slit positioned in the optical path for receiving light emitted from the light emitting end of the optical fiber and transmitting the light therethrough, a light scattering element disposed in the optical path wherein the light scattering element receives light transmitted through the entrance slit, separates the light into a plurality of light beams wherein each light beam has a different wavelength, and redirects the plurality of light beams back toward but offset from the entrance slit, and a light array detector capable of receiving the plurality of light beams and converting the plurality of light beams into an electrical signal for further processing.

The spectrometer module may have thermal compensation means for maintaining a position of the plurality of light beams in the light array detector. The thermal compensation means includes one or more insulators disposed around the spectrometer housing, a temperature controller assembly disposed in the spectrometer housing (the temperature controller assembly being, for example, a heating tape with a thermistor or other temperature measuring component and a program that controls heating of the tape as a function of the temperature within the spectrometer housing), and a thermal compensation lens mount.

The thermal compensation lens mount may have a fixed mount end and a non-fixed mount end that allows for thermal expansion and contraction of the thermal compensation lens mount. The fixed mount end is attached to a base plate or a bottom of the spectrometer housing. The lens mount has a coefficient of expansion greater than the coefficient of expansion of the base plate or spectrometer housing to which the lens mount is attached. The thermal compensation lens mount moves linearly and transversely relative to an optical path of light from the light entrance slit based on the coefficient of expansion of the lens mount. This temperature-based movement of the lens mount maintains the position of scattered light from the light scattering element to the light array detector. In other words, the thermal repositioning of the achromatic lens by the thermal compensation lens mount causes the scattered light from the light scattering element to strike the light array detector without affecting the electrical signal generated by the light array detector from the incident light. The shifting of the light beam is due to the light scattering element reacting to a change in temperature.

In another embodiment, the replaceable cuvette assembly may be installed in a compact spectrometer for measuring hemoglobin parameters in whole blood. The spectrometer includes a closed housing having a light input end/an optical fiber housing end with a light input port, a light input slit disposed on an electronic circuit substrate, the electronic circuit substrate disposed in the closed housing, where the light input slit is aligned with and adjacent to the light input port, a light array detector disposed on the circuit board substrate adjacent to the light input slit, and an optical component group formed by a light scattering element disposed downstream of the light input slit and a spherical achromatic lens disposed between the light input slit and the light scattering element, where the light scattering element has a reflective surface on a back side for reflecting scattered light toward the achromatic lens. The achromatic lens transmits light from the light inlet slit to the light scattering element and transmits the reflected scattered light from the light scattering element to the light array detector. To achieve this, the achromatic lens is slightly off-axis with respect to the light from the light inlet slit, so that the scattered light from the light scattering element is not directed back to the light inlet slit, but to the light array detector.

In a further example, a method of measuring hemoglobin parameters in whole blood despite the strong optical scattering caused by whole blood is disclosed. The method includes providing a light source such as an LED light source with a spectral range of about 422 nm to about 695 nm, guiding light having the spectral range of the light source along an optical path, providing a cuvette module with a sample receiving chamber having a first cuvette window disposed in the optical path where the first cuvette window transmits the light through the sample receiving chamber and through a second cuvette window aligned with the first cuvette window where the sample receiving chamber contains a whole blood sample, providing a pair of diffusers (i.e., a first diffuser and a second diffuser) disposed in the optical path where the first cuvette window and the second cuvette window of the sample receiving chamber of the cuvette are disposed between the pair of diffusers, guiding light from the cuvette module to a spectrometer having a light scattering element that separates the light into a plurality of light beams where each light beam has a different wavelength and convert ... plurality of light beams into an electrical signal, and processing the electrical signal into an output signal usable for displaying and reporting hemoglobin parameter values and/or total bilirubin parameter values of the whole blood sample.

In another example of the method, the processing step includes processing the electrical signal to spectral absorbance and then sweeping the spectral absorbance to hemoglobin parameter values and/or bilirubin parameter values using a computational sweep function.

In yet another example of the method, the processing step includes using a kernel-based orthogonal projection sweep function to latent structures as the computational sweep function.

In another example of the method, a method of measuring hemoglobin parameters in a whole blood sample is disclosed. The method includes (1) measuring and recording a scan of transmitted light intensity over a plurality of wavelengths in a measurement range by transmitting light through a cuvette module having an optical path with a known optical path length therethrough where the cuvette module is filled with a transparent fluid, (2) measuring and recording a scan of transmitted light intensity over the plurality of wavelengths of the measurement range by transmitting light through the cuvette a second time by having the optical path with the known optical path length therethrough where the cuvette module is filled with a whole blood sample, wherein each step of measuring and recording the transparent fluid and the whole blood sample includes scattering and circular polarization of the transmitted light prior to transmitting the transmitted light through the cuvette module and then scattering the transmitted light emitted by the cuvette module prior to determining a spectral absorbance, (3) determining a spectral absorbance at each wavelength of the plurality of wavelengths of the measurement range. of measurement based on a ratio of the transmitted light intensity scan of the whole blood sample to the transmitted light intensity scan of the clear fluid using a prism-based spectrometer, and (4) correlating the absorbance at each wavelength of the plurality of wavelengths in the measurement range with the hemoglobin parameter values and/or the bilirubin parameter values of the blood sample using a computational scanning function.

Brief description of the drawings

Figure 1 is a simplified perspective view showing a compact COOx subsystem comprising a replaceable cuvette assembly in accordance with one embodiment of the present invention.

Figure 2 is a side elevation view of one embodiment of the sample optical module shown in Figure 1. Figure 3 is a front perspective view of one embodiment of a light emitting module of the sample optical module shown in Figure 2.

Figure 3A is a front perspective view of the light emitting module shown in Figure 3 showing a plurality of optical components.

Figure 3B is an enlarged side elevation view of the optical components shown in Figure 3A. Figure 4 is a front perspective view of one embodiment of a cuvette assembly of the sample optical module shown in Figure 1.

Figure 5 is a rear perspective view of the bucket assembly shown in Figure 4.

Figure 6 is a front elevation view of a cuvette module of the cuvette assembly showing fluid inlet and outlet ports, a sample receiving chamber, a sample window, and an electronic chip assembly.

Figure 7 is a rear perspective view of the sample receiving chamber of Figure 6 showing the first and second cuvette windows.

Figure 8 is a rear plan view of the sample receiving chamber showing the electronic chip assembly disposed adjacent to the sample receiving chamber.

Figure 9 is a perspective view of one embodiment of a calibration light module of the sample optical module of Figure 1.

Figure 10 is a side cross-sectional view of the calibration light module of Figure 8 showing a calibration light source.

Figure 11 is a simplified side plan view of the calibration light source of the calibration light module of Figure 9 showing a plurality of optical components.

Figure 12 is a front perspective view of one embodiment of the spectrometer module of Figure 1 with a cover removed showing internal components.

Figure 13 is a rear perspective view of the spectrometer module of Figure 12 showing an input light slit and an adjacent light array detector.

Figure 14 is a rear cross-sectional view of the spectrometer module of Figure 12 showing a single circuit board and the location of the input light slit and the light array detector.

Figure 15 is a top view of the spectrometer module of Figure 12 showing the optical components with the ray trace superimposed.

Figure 16 is a ray trace showing the input light from the input light slit and a plurality of light beams refracted in the light array detector.

Figure 17A is a perspective view of one embodiment of a thermal compensation means for the spectrometer module showing insulation wrapped around the spectrometer module.

Figure 17B is a perspective view of another embodiment of a thermal compensation means for the spectrometer module showing a temperature control assembly.

Figure 17C is a cross-sectional view of one embodiment of a lens mount of the spectrometer module of Figure 12 showing a temperature compensated lens mount.

Figure 18 is a cross-sectional view of one embodiment of a lens mount of the spectrometer module of Figure 12 showing a fixed lens mount.

Figure 19 is a graphical illustration showing the correlation results of the COOx analyzer subsystem comprising a replaceable cuvette assembly of the present invention for total hemoglobin using a K-OPLS scanning function and method.

Figure 20 is a graphical illustration showing the correlation results of the COOx analyzer subsystem comprising a replaceable cuvette assembly of the present invention for oxyhemoglobin using a K-OPLS scanning function and method.

Figure 21 is a graphical illustration showing the correlation results of the COOx analyzer subsystem comprising a replaceable cuvette assembly of the present invention for carboxyhemoglobin using a K-OPLS scanning function and method.

Figure 22 is a graphical illustration showing the correlation results of the COOx analyzer subsystem comprising a replaceable cuvette assembly of the present invention for deoxyhemoglobin using a K-OPLS scanning function and method.

Figure 23 is a graphical illustration showing the correlation results of the COOx analyzer subsystem comprising a replaceable cuvette assembly of the present invention for methemoglobin using a K-OPLS scanning function and method.

Figure 24 is a graphical illustration showing the correlation results of the COOx analyzer subsystem comprising a replaceable cuvette assembly of the present invention for total bilirubin using a K-OPLS scanning function and method.

Detailed description

Embodiments of the present invention are illustrated in Figures 1-24. 1 shows one embodiment of a COOx analyzer subsystem 10. The COOx analyzer subsystem 10 includes at least one sample optical module 20, an optical fiber 90, and a spectrometer module 100. The COOx analyzer subsystem 10 may optionally include a processor module 150, or the processor module 150 may optionally be included in an electronic circuit of a diagnostic system of which the COOx analyzer subsystem 10 is a part. Line 5 is included to indicate that the processor module 150 may or may not be a part of the COOx analyzer subsystem 10. The processor module 150 includes, but is not limited to, a microprocessor module 152 and a memory module 154. Optionally, the processor module 150 may also include a converter module 156, or the converter module 156 may be external to the COOx analyzer subsystem 10. COOx 10. The COOx 10 analyzer subsystem is used to measure blood hemoglobin parameters such as total hemoglobin (tHb), carboxyhemoglobin (COHb), deoxyhemoglobin (HHb), oxyhemoglobin (O2Hb), methemoglobin (MetHb) and fetal hemoglobin (FHb) as well as total bilirubin (tBil) using optical absorbance.

Figure 2 illustrates sample optical module 20. Sample optical module 20 includes a light emitting module 22, a cuvette assembly 40, and a calibration light module 60. Light emitting module 22, as the term implies, emits a beam of visible light toward cuvette assembly 40 which is received by calibration light module 60, which is then transmitted to spectrometer module 100. Light beam 12 defines an optical path 21.

3-3A illustrate perspective views of one embodiment of the light emitting module 22 of FIG. 2. The light emitting module 22 includes a light emitting module substrate 24 containing an electrical circuit (not shown) and a light emitting optical assembly 25. The light emitting optical assembly 25 has an optical assembly housing 26 with an optical assembly end 26a. A visible light beam 28a is emitted from the optical assembly end 26a of the light emitting optical assembly 25 when the light emitting module 22 is turned on by a signal received from the processor module 150. FIG. 3A illustrates the light emitting optical assembly 25 with the optical assembly housing 26 removed exposing a plurality of optical components B contained within the light emitting assembly 25.

Turning now to Figure 3B, an enlarged side view of the plurality of optical components B of Figure 3A is illustrated. In this embodiment, the optical components B include a light emitting diode (LED) light source 28, a collimating lens 30, a first diffuser 32, a circular polarizer 34, a focusing lens 36, and an optional protection window 38. The circular polarizer 34 provides a distinct advantage. This advantage provides increased sensitivity and accuracy of the system. Hemoglobin has rotating optical characteristics, meaning that the polarization sensitivity of a spectrometer will cause an absorbance error if non-circularly polarized light is used to measure the absorbance of hemoglobin. Unlike other polarization states of light, the polarization state of circularly polarized light does not change as it passes through hemoglobin. Therefore, the polarization response of the spectrometer is the same for circularly polarized light passing through hemoglobin as for the reference scan taken with the cuvette filled with a transparent fluid.

4 and 5 illustrate front and rear perspective views of one embodiment of the cuvette assembly 40. The cuvette assembly 40 includes a cuvette substrate 41 and a cuvette module 43. The cuvette substrate 41 provides a support for securing the cuvette assembly 40 within the analyte subsystem 10 and includes a cuvette light path opening 42 that is disposed within the optical path 21 and aligned with the light beam emitted from the light emitting module 22. The cuvette module 43 includes a first cuvette portion 44 having a sample receiving cavity 45, a sample inlet port 46, a sample outlet port 47, an electronic chip assembly 48, and a first cuvette window 49, and a second cuvette portion 50 having a second cuvette window 52 (shown in FIG. 6 and outlined as outline 53) opposite and aligned with the first cuvette window 49 where the first and second cuvette windows 49, 52 are aligned and dispersed within optical path 21. The first cuvette portion 44 and the second cuvette portion 50 are bonded together with or without a gasket disposed between the first and second cuvette portions 44, 50. The bonding may be achieved by adhesives, ultrasonic techniques, solvent-based techniques, etc. Once assembled and as shown in Figure 6, the sample receiving cavity 45 of the first cuvette portion 44 forms a sample receiving chamber 54 with the second cuvette portion 50 fluidly communicating with the sample inlet and outlet ports 46, 47. The distance between the first and second cuvette windows 49, 52 of the sample receiving chamber 54 defines a cuvette optical path length, which is accurately measured and stored in the electronic chip 48 for later retrieval by the processor module 150. A typical optical path length used in this embodiment of the present invention is 0.0035 inches (0.090 mm).

Turning now to Figure 7, an enlarged rear perspective view of the first and second cuvette portions 44, 50 is illustrated. As shown, the first cuvette portion 44 has a sample chamber cavity 45 with a first cuvette window 49 and an electronic chip cavity 48a for receiving the electronic chip assembly 48. The second cuvette portion 50 has a second cuvette window 52 that forms the sample receiving chamber 54 when assembled together with the first cuvette portion 44. The second cuvette window 52, outlined by an outline 53 on the second cuvette portion 50, is a raised surface that forms a hermetic seal around the sample chamber cavity 45 and the sample receiving chamber 54. Optionally, a thin gasket may be placed between the first and second cuvette portions 44, 50 to more easily ensure a hermetic seal. 8 shows a rear view of the first cuvette portion 44 with the electronic chip assembly 48 disposed within the electronic chip cavity 48a. The electronic chip assembly 48 includes a chip circuit board 48b and an electronic chip 48c that stores the cuvette optical path length value for the particular cuvette module 43. The first cuvette window 49 is disposed within the optical path 21 and transmits the light beam passing through the sample to the calibration light module 60, which then passes the light beam to the spectrometer module 100.

Turning now to Figure 9, one embodiment of the calibration light module 60 is illustrated. The calibration light module 60 includes a calibration module housing 62, a light beam receiving portion 64, a calibration light portion 70, and an optical fiber portion 80 wherein the calibration module 62 is received, the light beam receiving portion 64 and the optical fiber portion 80 are aligned with the optical path 21. The calibration light portion 70 is spaced apart from and transverse to the optical path 21.

Figure 10 is a cross-section, elevation view of calibration light module 60. Calibration module housing 62 includes a first tubular conduit 62a between a light beam input opening 62b and a light beam output opening 62c, as well as a second tubular conduit 62d that is transverse to and intersects first tubular conduit 62a at one end and has a calibration light beam opening 62e at an opposite end.

The light beam receiving portion 64 houses a collimating lens 66 that collimates the light beam 28a received along the optical path 21 from the cuvette module 43 and directs the light beam 28a into the first tubular conduit 62a. Disposed within the calibration module housing 62 is the beam splitter support assembly 67 which is disposed transversely across the first tubular conduit 62a. The beam splitter support assembly 67 has an upwardly inclined surface 67a facing the calibration light beam aperture 62e and the light beam exit aperture 62c within the optical path 21. The beam splitter support assembly 67 supports a second diffuser 68 and a beam splitter 69 (shown in Figure 11) that is disposed downstream along the optical path 21 from the second diffuser 68 such that it is positioned to receive the calibration light beam 72a and direct it along the optical path 21 and the first tubular conduit 62a toward the light beam exit aperture 62c.

The calibration light portion 70 includes a calibration light source 72 disposed adjacent to but spaced apart from the optical path 21 that is capable of directing a calibration light beam 72a into the calibration module housing 62 through a calibration light aperture 62e transversely to the optical path 21 toward the beam splitter support assembly 67. Within the calibration light portion 70, there is a collimating lens 74 that collimates the calibration light beam 72a before it is reflected by the beam splitter assembly 67 toward the light beam exit aperture 62c.

The optical fiber portion 80 is positioned within the optical path 21 at or near the light beam exit aperture 62c. The optical fiber portion 80 includes a focusing lens 82 and a fiber optic connector assembly 84 including a connector housing 86 adapted to receive an optical fiber assembly 90. The optical fiber portion 80 is adapted to ensure that the light beam 28a is properly focused by the focusing lens 82 onto the optical fiber assembly 90.

Figure 11 is a simplified illustration of Figure 10 showing the positional relationship of optical components 66, 68, 69, 74, 82 and light beams 28a, 72a as well as fiber optic assembly 90. As can be seen in Figure 11, light beam 28a is received by collimating lens 66, transmitted through second diffuser 68 and beam splitter 69 to focusing lens 82 and fiber optic assembly 90. As discussed above, the importance of using a pair of diffusers (first diffuser 32 and second diffuser 68) with cuvette module 43 between the pair of diffusers 32, 68 is that the spatial light distribution will appear the same for the blank scan and the whole blood sample scan. The use of diffusers 32, 68 in this arrangement eliminates the effect of error caused by non-uniformity of the light source and/or variation in spatial distribution changes of the incident light, even if the overall intensity has not changed. The diffusers 32, 68 are chosen so as to spread an incident light ray into the full acceptance cone of the optical component array 120 of the spectrometer module 100. In this way, the ray completely traverses the optical measurement field.

The calibration light beam 72a, when activated, is received by the collimating lens 74, transmitted to the beam splitter 69, and directed to the focusing lens 82 where it is focused onto the optical fiber assembly 90. The calibration light beam 72a has specific wavelengths of light used to calibrate the wavelength scale of the spectrometer module 100. An example of an acceptable calibration light source 72 is a krypton (Kr) gas discharge lamp, which provides seven Kr line wavelengths in nanometers covering the range 422 to 695 nm. The prism 131 of the light dispersing component 130 has a non-linear dispersion as a function of wavelength requiring a polynomial or other higher order function. The present disclosure uses a 5th order polynomial for the Kr line peak pixel locations to provide residual errors well below the absolute wavelength accuracy requirement of ± 0.03 nm.

The optical fiber assembly 90 comprises an optical fiber 92, a first optical fiber connector 94, and a second optical fiber connector 96 (shown in Figure 12). The first optical fiber connector 94 is secured to a light receiving end 92a of the optical fiber 92 and is directly and removably connected to the connector housing 86 of the optical fiber connector assembly 84. One embodiment of the optical fiber 92 includes a 200 μm silica core fiber with a numerical aperture (NA) of 0.22.

Turning now to Figures 12 and 13, one embodiment of the spectrometer module 100 is illustrated. The spectrometer module 100 includes a spectrometer housing 102, a spectrometer base 104, a spectrometer cover 106 (shown in Figure 1), an optical fiber housing end 108, and an electrical signal output coupler 103. The spectrometer module 100 has outer envelope dimensions of 11 cm x 8 cm x 2 cm and optionally includes thermal compensation structures discussed below. Contained within the spectrometer housing 102 are the essential components of the spectrometer module 100. These components include a light receiving and converting assembly 110 and an optical component assembly 120. The optical component assembly 120 includes an achromatic lens assembly 121 and a light dispersing element 130. The light dispersing element 130 may be a prism 131 or a grating 136. The optical fiber assembly 90 is removably secured to the optical fiber housing end 108 at the light input port 109, which optical fiber assembly 90 transmits light beams 28a, 72a to the spectrometer module 100. As mentioned above, light beam 28a represents the light transmitted from the light emitting module 22 through the cuvette module 43, while light beam 72a is the light emitting module 22. calibration light transmitted from the calibration light module 60, which is used to calibrate the spectrometer module 100.

The achromatic lens assembly 121 includes a lens mount 122 and a spherical achromatic lens 124. The achromatic lens 124 receives the light beams 28a, 72a, as the case may be, and directs the light beam toward the light dispersing element 130, which in this embodiment is the prism 131. The prism 131 has a reflective coating 132 on an exterior back surface. The prism 130 refracts the light beam 28a and reflects the light back through the achromatic lens 124.

The light receiver and converter assembly 110 is securedly mounted adjacent an interior surface 108a of the optical fiber housing end 108. The light receiver and converter assembly 110 includes a circuit board substrate 112 on which is mounted a light input slit 114 that is aligned with the light emitting end 92b (not shown) of the optical fiber 92. Adjacent to the input slit 114 is a light array detector 116 that receives refracted light from the prism 131. The light array detector 116 converts the refracted light into an electrical signal, which is sent through the output connector 118 to the processor module 150. Arranging the light input slit 114 and the light array detector 116 adjacent to each other on the circuit board 112 has several advantages. This feature greatly simplifies the construction and improves the accuracy of the spectrometer module 100. Other spectrometers place these elements on separate planes, where they have separate mounting structures and must be independently adjusted. This feature of mounting the entrance slit and the light array detector adjacent to each other on the circuit board 112 eliminates the need to mount and position each structure (i.e., slit and detector) separately.

14 is an enlarged view of the light receiver and converter assembly 110. The light entrance slit 114 is 15 µm wide by 1000 µm long which projects a fiber optic slit image that is a rectangle about 15 µm wide by 200 µm high onto the light array detector 116 (Hamamatsu S10226-10 is an example of a light array detector that can be used). The entrance slit 114 is applied directly onto the same circuit board substrate 112 and in close proximity to the light array detector 116. The light array detector 116 has a pixel height between about 100 to about 150 µm, which allows for one-to-one imaging of the 200 µm diameter optical fiber onto the detector. In this embodiment, the entrance slit 114 is laser etched at a precise position relative to the light array detector 116, making alignment less laborious. Because the entrance slit 114 and the light array detector 116 are only slightly offset from the axis relative to the central axis of the achromatic lens 124, there is minimal aberration and one-to-one imaging is possible at the light array detector 116, so that a cylindrical focusing lens is not needed to down-image the optical fiber (200 µm diameter fiber) to match the pixel height of the light array detector 116.

Turning now to Figure 15, there is a top view of the spectrometer module 100 of Figure 13. Superimposed on Figure 15 is a ray trace diagram 140 of the light beam supplied to the spectrometer module 100 by the optical fiber 92. As shown, the light beam 28a enters the spectrometer module 100 through the entrance slit 114 into the achromatic lens 124. The achromatic lens 124 is used off-axis; that is, the achromatic lens is slightly offset from the axis of the light beam 28a. The light beam 28a is transmitted by the achromatic lens 124 to the prism 131, where the light beam 28a is refracted into a plurality of light beams 138a, 138b, 138c of different wavelengths, as prisms should do. The plurality of light beams 138a, 138b, 138c are reflected by the prism 131 through the achromatic lens 124. The achromatic lens 124 is used off-axis to direct the plurality of refracted and reflected light beams 138a, 138b, 138c from the prism 131 to the light array detector 116.

Figure 16 is an enlarged view of ray tracing diagram 140. Achromatic lens 124 is used off-axis with respect to the incoming light beam 28a. By utilizing off-axis achromatic lens 124 in conjunction with prism 131 having a reflective coating 132 on a base of prism 131, a compact, simplified, minimal component spectrometer module 100 is achieved capable of being used to measure hemoglobin parameters and/or total bilirubin parameters in whole blood.

A change in temperature has a greater effect on the angle of refraction of the beam when a prism is used rather than a diffraction grating. A thermal compensation means 160 is provided to compensate for a thermal shift in the incoming light beam caused by the light dispersing element 130. A temperature change within the spectrometer module 100 causes a thermally induced movement of the image of the entrance slit 114 in the light array detector 116 caused in turn by thermally induced changes in the refractive index of the dispersive prism 131. Figure 16 shows the direction of movement of the image in the light array detector 116 for the thermal change in refractive index in the prism 131 with arrow 400. If the lens 124 is moved in the opposite direction over the same temperature range, as indicated by arrow 402, the image of the slit will move back to where it should be in the light array detector 116. To prevent this shift, the thermal compensation means 160 may be as simple as wrapping the module in a plastic bag. 17A and 17B illustrate these possibilities.

In one embodiment shown in Figure 17C, the achromatic lens mount 122 is a thermally compensating lens mount. The thermally compensating lens mount 122 has a fixed mounting end 122a and a non-fixed mounting end 122b. The fixed mounting end 122a is fixedly secured to the spectrometer base 104 or to a base plate 104a that is securely attached to the spectrometer base 104. The non-fixed mounting end 122b typically has a clamping element 126 that extends through a lens mounting slot 122c of the lens mount 122 and into the spectrometer base 104 or the base plate 104a. Between a head 126a of the clamping member 126 and the lens mount 122 there is a retaining spring 128. There is sufficient space between the lens mount slot 122c and the clamping member 126 to allow for expansion/contraction of the lens mount 122 caused by a temperature change. The coefficient of expansion of the lens mount 122 is greater than the coefficient of expansion of the spectrometer base 104 and/or the base plate 104a, so that the non-fixed mount end 122b allows thermal expansion and contraction of the thermally compensated lens mount 122 in a direction indicated by arrow 500, which is linear and transverse to the light beam from the entrance slit 114. This structure allows the achromatic lens 124 to slide relative to other components mounted on the base plate 104a and/or the spectrometer base 104. The thermally compensated lens mount 122 ensures that the plurality of light beams 138a, 138b, 138c will always strike the light array detector 116 with sufficient intensity without affecting the electrical signal generated by the light array detector 116 despite a change in the direction of the light array detector 116. temperature within the spectrometer housing 102. One such material that meets the requirement that the lens mount 122 have a greater coefficient of expansion than the spectrometer base 104 and/or base plate 104a (as the case may be) is a plastic that is a modified polyphenylene ether (PPE) resin consisting of amorphous mixtures of polyphenylene ether (PPE) resin, polyphenylene oxide (PPO), and polystyrene that is sold under the trademark NORYL®.

18 illustrates an alternative embodiment of the lens mount 122. In this embodiment, the lens mount 122 has two fixed mount ends 122a, where each end 122a is secured to the base plate 104a and/or the spectrometer base 104 by the clamping element 126. Because both ends 122a of the lens mount 122 are fixed, any temperature change within the spectrometer module 100 will affect the angle of the plurality of light beams 138a, 138b, 138c and where they strike the light array detector 116. As noted above in connection with the slit image and the length of the light array detector 116, a temperature change greater than 0.5°C will cause the intensity of one of the light beams to not fully strike the light array detector, thereby causing a low reading. inaccurate. To negate this potential effect, the spectrometer module 100 is equipped with a temperature controller assembly (not shown) so that the prism 131 and achromatic lens assembly 121 remain at a constant temperature. Although there are various methods for maintaining the interior of the spectrometer module 100 at a constant temperature, one example of such a temperature controller assembly to accomplish this is a ribbon heater with a thermistor (not shown) adhesively attached to the interior or exterior of the spectrometer module 100, such ribbon heater being controlled by an electronic regulating circuit (not shown). Optionally, the spectrometer module 100 may also be insulated inside or out, or both, to more easily maintain a given temperature and protect against temperature changes in the vicinity surrounding the spectrometer module 100. Other mechanisms include placing the spectrometer module 100 within a temperature controlled environment.

Learning data:

A dataset of approximately 180 blood samples from approximately 15 different individuals was developed. Blood samples were manipulated using sodium nitrite to elevate MetHb values and using CO gas to elevate COHb values. Plasma was extracted from or added to the samples to modify the tHb level. To vary the tBil level, a bilirubin solution was added. A tonometer was used to manipulate the oxygen level. Blood samples were manipulated to cover a wide range of analyte values. Blood samples were then measured on a reference lysis pHOx Ultra analyzer equipped with COOx analyzer and analysis software. Whole blood spectra were collected on a pHOx Ultra analyzer equipped with the high angle collection optics and other modifications of the present invention as described above, with the lysate supply line completely disconnected and whole blood samples passed directly into cuvette assembly 40 without lysate or any further dilution. Both analyzers were equipped with Zeonex windows in the respective cuvettes. This data set has been converted to a Matlab cell array file for use with Matlab instruction sets.

Prediction model:

The next stage of the calculation is to create a prediction model. Three models were developed for the analysis: one for the COOx parameters tHb and COHb, a second for HHb and MetHb, and a third for tBil. The amount of O2Hb was determined by subtracting that of COHb, HHb, and MetHb from 100%. The data matrix X was constructed from terms created from the absorbance measured at wavelengths between 462-650 nm, 1 nm spacing. The tBil model was developed using the same data set as the COOx model, except that samples with MetHb values greater than or equal to 20% were left out of the model. For each model, five predictive Y values (O2Hb, HHb, COHb, MetHb, tBil) were assigned with tHb determined by summing the results for O2Hb, HHb, COHb, and MetHb. The number of orthogonal Y values required was determined by manually optimizing the correlation residual of the blood predictions from the sweep function with the reference analyzer values.

Using an initial calibration data set, the calibration sequence of a machine learning algorithm establishes a relationship between a matrix of known sample characteristics (the Y matrix) and a matrix of measured absorbance values at various wavelengths and potentially other measured values based on absorbance versus wavelength (the X matrix). Once this relationship is established, it is used by the analyzer to predict unknown Y values from new X measurements on whole blood samples.

The fits and inputs used for the optimized models are summarized in Table 1. The X data consist of absorbance and other absorbance-versus-wavelength-based terms. In the process of model optimization, absorbance-versus-wavelength derivatives were added. Models for analytes most sensitive to nonlinear scattering effects were built with square root terms of absorbance and its derivative. The model for analytes most affected by scattering had a correction term proportional to the fourth power of the wavelength. The X vector row has a value for each wavelength for each of the three absorbance-based terms f, g, and h shown in the table for each model.

�� The calibration set matrix Y is constructed as follows from the known values of the calibration sample set of n lysed blood samples:

where

tHb is the total hemoglobin value of the lysed blood sample,

COHb is the carboxyhemoglobin value of the lysed blood sample,

HHb is the deoxyhemoglobin value of the lysed blood sample,

MetHb is the methemoglobin value of the lysed blood sample and

tBil is the total bilirubin value of the lysed blood sample.

The matrix X is structured as follows:

where: f, g, h are the absorbance-based functions listed in Table 1 as a function of wavelength, respectively.

The matrixX includes the absorbance contributions at the various wavelengths. The disclosure optionally includes the addition of other measurements to the calculation to reduce interfering effects.

Once these matrices are formed, they are used as a calibration set and the assignment function is calculated according to the procedures of the chosen machine learning algorithm.

As previously described, conventional partial least squares, linear regression, linear algebra, neural networks, multivariate adaptive regression splines, projection to latent structures, kernel-based orthogonal projection to latent structures, or other machine learning mathematics is used with the results obtained from the calibration data set to determine the empirical relationship (or sweep function) between absorbance values and hemoglobin parameters. Typically, a mathematics package is used to generate the results, where the package usually has options to select one of the machine learning mathematics known to those skilled in the art. A variety of mathematics packages exist and include, but are not limited to, Matlab from MatWorks of Natick, MA, "R" from the R Project for Statistical Computing available on the Internet at www.r-project.org, Python from the Python Software Foundation and available on the Internet at www.python.org in combination with Orange data mining software from Orange Bioinformatics available on the Internet at orange.biolab.si, to name a few.

It will be demonstrated that the kernel-based orthogonal projections to latent structures (KOPLS) method can be used as a kind of machine learning algorithm to generate the assignment function. KOPLS is best explained and described by the following references: Johan Trygg and Svante Wold, "Orthogonal projections to latent structures (O-PLS)", J. Chemometrics 2002; 16: 119-128; Mattias Rantalaine et al. "Kernel-based orthogonal projections to latent structures (K-OPLS)", J. Chemometrics 2007; 21: 376-385; and Max Bylesjoet al. "K-OPLS package: Kernel-based orthogonal projections to latent structures for prediction and interpretation in feature space", BMC Bioinformatics 2008, 9:106. Kernel-based mathematics is useful for dealing with nonlinear behavior of systems by using a kernel function to map the original data to a higher-order space. While any of the machine learning mathematics described above can be used, KOPLS has an additional advantage over other calculations such as conventional partial least squares because it can not only establish a relationship between the quantified variations and the analyte values to be determined, but it can also remove the unquantified variation that is always present in the original data. These unquantified variations may be due to analyzer and/or blood effects such as scattering losses and other interference phenomena that are not explicitly measured. By extracting these unquantified variations from the data, the method leaves in the data the information used to predict the measured values.

Using an initial training data set, the KOPLS model establishes a relationship (sweep function) between the known feature matrix of the sample (the H matrix) and a matrix of measured absorbance values at various wavelengths and potentially other measured values based on absorbance versus wavelength (the X matrix), as processed through a kernel function specified by the KOPLS method. Once the KOPLS coefficients of this relationship are established, they are used with the kernel function by the analyzer to predict unknown values of hemoglobin parameters from new absorbance measurements on the samples.

The kernel function used in this example is a simple linear kernel function described in the reference Mattias Rantalainenet et al. mentioned above and represented by the following equation:

where the matrix of measured values X is fed into the kernel function and subjected to further processing as specified in the KOPLS references cited above to create the KOPLS training coefficients.

Once the set of training coefficients, or sweep function, is established, it is used to predict the hemoglobin and/or total bilirubin parameter values of a blood sample from future measurements. A single-row matrix X is created from the new measurements, the value of this single-row matrix X is then passed through the kernel and sweep functions to produce the hemoglobin parameter values and/or the total bilirubin parameter values according to the procedures required for the sweep function used in accordance with the KOPLS procedures described in detail in the previously disclosed KOPLS references.

The data collected from the blood samples described above were subjected to the KOPLS method in a cross-validation process. Cross-validation is a process of using a data set to test a method. Several rows of data are held out and the remainder is used to create an assignment function. The removed values are then used as "new" measurements and their Y matrix values are calculated. This process is repeated by setting aside other measured values and calculating another assignment function. By plotting the known values of the blood data against the calculated ones, the effectiveness of the method can be checked by inspecting the plot.

Moving on to Figures 18-23, the graphs of the correlation results comparing the various hemoglobin parameters of lysed blood with whole blood using the KOPLS method are illustrated. The blood samples were manipulated to cover a wide range of analyte values. To test the data, the n-fold cross-validation technique was used using 60 partitions. In this technique, the data set is divided into n=60 independent sets and the model is built from n-1 of the sets, with the remaining set being predicted using the model. The process is repeated 60 times for each group. Thus, each data point is predicted using a model built from most of the other data points, without being included in the model.

Figure 19 shows the correlation results for tHb using the K-OPLS method. The horizontal axis has units representing total hemoglobin in grams per deciliter of lysed blood. The vertical axis has units representing total hemoglobin in grams per deciliter of whole blood. As can be seen from the diagram, the method of determining tHb from a whole blood sample has a correlation greater than 99%.

Figure 20 shows the correlation results for O2Hb using the K-OPLS method. The horizontal axis has units representing the percentage of oxyhemoglobin in lysed blood. The vertical axis has a unit representing the percentage of oxyhemoglobin in whole blood. As seen in the plot, the method of determining O2Hb from a whole blood sample has a correlation greater than 99%.

Figure 21 shows the correlation results for carboxyhemoglobin using the K-OPLS method. The horizontal axis has units representing the percentage of carboxyhemoglobin in lysed blood. The vertical axis has the unit representing the percentage of carboxyhemoglobin in whole blood. As seen in the plot, the method of determining COHb from a whole blood sample has a correlation greater than 99%.

Figure 22 shows the correlation results for deoxyhemoglobin using the K-OPLS method. The horizontal axis has units representing the percentage of deoxyhemoglobin in lysed blood. The vertical axis has a unit representing the percentage of deoxyhemoglobin in whole blood. As seen in the plot, the method of determining HHb from a whole blood sample has a correlation greater than 99%.

Figure 23 shows the correlation results for methemoglobin using the K-OPLS method. The horizontal axis has units representing the percentage of methemoglobin in lysed blood. The vertical axis has the unit representing the percentage of methemoglobin in whole blood. As seen in the plot, the method of determining MetHb from a whole blood sample has a correlation greater than 99%.

Figure 24 shows the correlation results for tBil using the K-OPLS method. The horizontal axis has units representing total bilirubin in milligrams per deciliter of lysed blood. The vertical axis has units representing total bilirubin in milligrams per deciliter of whole blood. As can be seen from the diagram, the method of determining tBil from a whole blood sample has a correlation greater than 99%.

Next, a method for performing a whole blood measurement using the COOx analyzer subsystem 10 of the present invention will be described. An absorbance scan is measured by first recording a transmitted light intensity scan with the cuvette module 43 filled with a clear fluid such as water or analyzer wash solution, also known as a "blank" scan. Next, a transmitted light intensity scan is recorded with the cuvette module 43 filled with the whole blood sample. After corrections for spectrometer dark response and detector linearity, the spectral absorbance is the negative of the base ten logarithm of the ratio of the whole blood scan to the clear fluid scan calculated at each wavelength in the measurement range.

More specifically, a representation of the components of a COOx analyzer subsystem is shown in Figures 1-18. This embodiment of the subsystem measures the optical absorbance of liquids introduced into the cuvette module 43. The light used to make the absorbance measurement originates from the LED light source 28, is collected and transmitted by the collimating lens 30, passes through the first diffuser 32, circular polarizer 34, focusing lens 36, and optional protective window 38 before reaching the cuvette module 43. Knowledge of the cuvette path length is critical to an absolute absorbance measurement. The cuvette path length is pre-measured for each individual cuvette module 43 and is programmed into an electronic chip 48c in the cuvette module 43. The data processor module 130 of the analyzer reads/retrieves the path length information whenever needed.

After passing through the cuvette module 43, the light is collected by lens 66, collimated, and sent through second diffuser 68 and beam splitter 69. The purpose of beam splitter 69 is to allow light from calibration light source 72 (e.g., a krypton gas discharge lamp), collimated by lens 74, to enter optical path 21. Calibration light source 72 provides light at known wavelengths, which are used to periodically recalibrate the wavelength scale of spectrometer module 100. After passing through beam splitter 69, the light is focused by lens 82 onto an optical fiber 92. Optical fiber 92 guides the light to entrance slit 114 of spectrometer module 100. Light passes through an achromatic lens 124, light dispersing element 130 with a backlight 140, and then passes through a dispersing element 130. The light is dispersed into wavelengths by passing through the light dispersing element 130, such as, for example, the prism 130 back through the lens 124, which refocuses the light onto the pixels of the light array detector 116. The light array detector 116 converts the light energy into an electrical signal representing the spectral intensity of the light. The electrical signal is sent to the data processor module 150 for further processing and display of the final results to the user. The light receiver and converter assembly 110 is a single board containing the entrance slit 114 and the light array detector 116 in close proximity as an integrated unit.

The entrance slit 114 is applied directly onto the same circuit board substrate 112 and in close proximity to the light array detector 116. Other prior art spectrometers place these components on separate planes where they have separate mounting structures that require separate adjustment and alignment. The mounting scheme of the present invention has several advantages that reduce the cost and size of the spectrometer module 100: 1) the cost of separate mounting structures is avoided, 2) the entrance slit 114 can be laser etched in a precise position relative to the light array detector 116 making alignment less labor intensive, 3) low cost spherical surface optics can be used in the optical system since the image of the slit on the detector is only slightly offset from the central axis of the optical system minimizing aberration, and 4) a single alignment procedure for a unified slit and detector assembly replaces the alignment procedures for two separate assemblies.

It is important to note that the first diffuser 32 and the second diffuser 68 are located before and after the cuvette module 43, respectively. Optical absorbance measurement of a diffuse sample presents a unique problem. The diffuse transmittance of the sample messes up the initial spatial light distribution of the measurement system caused by the typical non-uniformity of light sources. Therefore, the spatial light distribution of the "blank" scan can be very different from the whole blood sample scan. Since optical detectors have a spatially varying response, the response can vary due to changes in the spatial distribution of the incident light, even though the overall intensity has not changed. An absorbance scan based on the ratio of the sample scan to the blank scan will have a significant absorbance component due to this effect, in addition to the absorbance due to the sample alone. This results in a significant measurement error of the sample absorbance that is intolerable for co-oximetry.

The advantage of placing the cuvette module 43 between the first and second diffusers 32, 68 is that the spatial light distribution will remain the same for both the blank and sample scans, eliminating this error effect. The diffusers 32, 68 are specially chosen to spread an incident light beam over the entire acceptance cone of the optical system, but no more, so as to retain the highest possible luminous flux, while dispersing the light beam over the entire field.

Although preferred embodiments of the present invention have been described herein, the foregoing description is merely illustrative. Additional modifications of the invention disclosed herein will occur to those skilled in the respective arts and all such modifications are considered to be within the scope of the invention as defined in the appended claims.

Claims (7)

1. A replaceable cuvette assembly (40) capable of being installed in an optical absorbance measurement system for measuring hemoglobin parameters in whole blood or bilirubin parameters in whole blood, the cuvette assembly (40) comprising: a cuvette substrate (41); and a cuvette module (43) replaceably connected to the cuvette substrate (41), wherein the cuvette substrate (41) is a support configured to secure the cuvette assembly (40) within the optical absorbance measurement system, comprising the bucket module (43): a sample inlet port (46); a sample outlet port (47); a sample receiving chamber (54) fluidly communicating with the sample inlet port (46) and the sample outlet port (47); an electronic chip assembly (48) comprising a printed circuit board (48b) and an electronic chip (48c) with a sample receiving chamber cuvette optical path length value (54) programmed into the electronic chip (48c); a first bucket window (49); and a second cuvette window (52) forming a portion of the sample receiving chamber (54), wherein the first cuvette window (49) and the second cuvette window (52) are aligned with each other, thereby defining a cuvette optical path length having the cuvette optical path length value between the first cuvette window (49) and the second cuvette window (52), wherein the cuvette module (43) includes a first cuvette portion (44) and a second cuvette portion (50) joined together and thus forming the sample receiving chamber (54), the first cuvette portion (44) having an electronic chip cavity (48a), the electronic chip assembly (48) being disposed within the electronic chip cavity (48a), and wherein the first cuvette window (49) and the second cuvette window (52) are disposed within an optical path of the optical absorbance measurement system. 2. The replaceable cuvette assembly (40) of claim 1, wherein the cuvette substrate (41) has a cuvette light path opening (42) through the cuvette substrate (41), wherein the cuvette light path opening (42) is disposed within the optical path. 3. The replaceable bowl assembly (40) of claim 1 or 2, further comprising a gasket disposed between the first bowl portion (44) and the second bowl portion (50). 4. The replaceable cuvette assembly (40) of any one of the preceding claims, wherein the defined cuvette optical path length is accurately measured and a measurement result is stored as the cuvette optical path length value within the electronic chip (48c). 5. The replaceable cuvette assembly (40) of any one of the preceding claims, wherein the first cuvette portion (44) contains a sample receiving cavity (45), the sample inlet port (46), the sample outlet port (47), the electronic chip (48c) and the first cuvette window (49) and wherein the second cuvette portion (50) contains the second cuvette window (52). 6. The replaceable cuvette assembly (40) of any one of the preceding claims, wherein the second cuvette window (52) of the second cuvette portion (50) is a raised surface that forms a hermetic seal around a sample receiving cavity (45) of the first cuvette portion (44). 7. The replaceable cuvette assembly (40) of claim 1, wherein the cuvette module (43) is for use in an optical absorbance measurement system for measuring hemoglobin parameters in whole blood or bilirubin parameters in whole blood with the cuvette module (43) between a first optical scatterer (32) and a second optical scatterer (68) within the optical absorbance measurement system.